正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

It's never been a better time to be creative. (ever) The two best frontier models (ever) have been released within 30 days of each other. Here’s what we learned from running 4 frontier models head to head. >Sol has taste and Fable takes direction. 10 identical landing page briefs,... show more

ben

15,105 subscribers

35,609 次观看 • 2 个月前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

Fable is back (enjoy the final hours of in-sub usage) so we made it fight Anthropic's Opus 4.8: >5 real landing page and portfolio briefs, built in Claude Code, >judged blind by 9 working designers across 90 matchups. Fable's best brief won 88.9% of its head-to-heads. Its worst output only won 11.1%. The difference was the brief 🧵

Fable is back (enjoy the final hours of in-sub usage) so we made it fight Anthropic's Opus 4.8: >5 real landing page and portfolio briefs, built in Claude Code, >judged blind by 9 working designers across 90 matchups. Fable's best brief won 88.9% of its head-to-heads. Its worst output only won 11.1%. The difference was the brief 🧵

ben

105,014 次观看 • 2 个月前

🚨IT'S REAL: I’m a former HYBE employee, and I can affirm that they are a real couple and that their focus has been on each other ever since. I’m very afraid of being exposed and doxxed, but I was there in 2017 when their relationship was burgeoning, and in 2018, when someone found an opportunity to get involved and has been a nuisance ever since. As you can see, their connection has always been incredibly strong and real. They have been each other’s person from the very beginning, and that has never changed

Sensitive content

🚨IT'S REAL: I’m a former HYBE employee, and I can affirm that they are a real couple and that their focus has been on each other ever since. I’m very afraid of being exposed and doxxed, but I was there in 2017 when their relationship was burgeoning, and in 2018, when someone found an opportunity to get involved and has been a nuisance ever since. As you can see, their connection has always been incredibly strong and real. They have been each other’s person from the very beginning, and that has never changed

ᴍꜱʙᴀᴅꜱᴘᴇʟʟᴇʀ ♡ ⓋⓀ𐤀

826,941 次观看 • 14 天前

Grok 4.5 performed GPT Sol level for free! We gave 4 models the same prompt: build three self-contained HTML5 canvas scenes with real physics demos Prompts: -robot deathmatch, Tombstone vs Minotaur -a hydraulic press flattening stuff on a conveyor -a semi truck jumping a canyon Outputs: GPT-5.6 Sol: 12.9K tokens, $0.51 (~7 min) Grok 4.5: 10.8K tokens, $0 (~5 min) Muse Spark 1.1: 26.8K tokens, $0.12 (~7.5 min) GLM 5.2: 10.9K tokens, $0.02 (~12 min) Grok 4.5 handled all three scenes genuinely well and got surprisingly close to GPT-5.6 this round. On top of that, it ran on the free tier. GPT-5.6 Sol, the frontier model, put out solid but not standout work. GLM 5.2 rendered all three scenes for pennies, but it came out the roughest of the four. Meta's new Muse Spark burned the most tokens yet still stayed cheap, delivering an average result.

Grok 4.5 performed GPT Sol level for free! We gave 4 models the same prompt: build three self-contained HTML5 canvas scenes with real physics demos Prompts: -robot deathmatch, Tombstone vs Minotaur -a hydraulic press flattening stuff on a conveyor -a semi truck jumping a canyon Outputs: GPT-5.6 Sol: 12.9K tokens, $0.51 (~7 min) Grok 4.5: 10.8K tokens, $0 (~5 min) Muse Spark 1.1: 26.8K tokens, $0.12 (~7.5 min) GLM 5.2: 10.9K tokens, $0.02 (~12 min) Grok 4.5 handled all three scenes genuinely well and got surprisingly close to GPT-5.6 this round. On top of that, it ran on the free tier. GPT-5.6 Sol, the frontier model, put out solid but not standout work. GLM 5.2 rendered all three scenes for pennies, but it came out the roughest of the four. Meta's new Muse Spark burned the most tokens yet still stayed cheap, delivering an average result.

atomic.chat

70,490 次观看 • 2 个月前

Nearly a year of Code Arena: WebDev progress compressed into 15 seconds. Each line follows the highest-scoring model from top labs over time, showing the pace of improvement across the ecosystem. In just the last year, the model in the leading spot increased score by +340 pts and the number of frontier labs competing for the top spot expanded from 6 to 10. Anthropic has dominated throughout the year. Although standout releases have jumped to the top spot, most notably the Chinese open-source model Kimi K3 from Kimi.ai in July. Today, GPT-6 Astra by OpenAI leads with 1796 pts, followed by Claude Fable 5.1 at 1764 pts. The next closest lab is 103 pts away, Qwen with 1685 pts. Code Arena: WebDev ranks models through head-to-head user preference on real front-end web development tasks. These votes drive the leaderboard that is tracking the frontier.

Nearly a year of Code Arena: WebDev progress compressed into 15 seconds. Each line follows the highest-scoring model from top labs over time, showing the pace of improvement across the ecosystem. In just the last year, the model in the leading spot increased score by +340 pts and the number of frontier labs competing for the top spot expanded from 6 to 10. Anthropic has dominated throughout the year. Although standout releases have jumped to the top spot, most notably the Chinese open-source model Kimi K3 from Kimi.ai in July. Today, GPT-6 Astra by OpenAI leads with 1796 pts, followed by Claude Fable 5.1 at 1764 pts. The next closest lab is 103 pts away, Qwen with 1685 pts. Code Arena: WebDev ranks models through head-to-head user preference on real front-end web development tasks. These votes drive the leaderboard that is tracking the frontier.

Arena.ai

93,691 次观看 • 21 天前

How much better are the internal, unreleased models at frontier labs like Google, OpenAI, and Anthropic? We got a glimpse exactly one year ago today, when Google accidentally leaked the “Kingfall” model "Kingfall" was likely an unreleased Gemini 2.5 Ultra-sized model. It was available in AI Studio for only a few minutes but remained accessible through the API for several days At the time, "Kingfall" appeared to be significantly better than Gemini 2.5 Pro at both code generation and creative writing In a recent interview, Sundar Pichai mentioned that Google could have made a better, Ultra-sized Gemini Omni model, but would have had trouble serving it The infrastructure required to serve Ultra-sized models at scale is likely why Google never publicly released models like “Kingfall”

How much better are the internal, unreleased models at frontier labs like Google, OpenAI, and Anthropic? We got a glimpse exactly one year ago today, when Google accidentally leaked the “Kingfall” model "Kingfall" was likely an unreleased Gemini 2.5 Ultra-sized model. It was available in AI Studio for only a few minutes but remained accessible through the API for several days At the time, "Kingfall" appeared to be significantly better than Gemini 2.5 Pro at both code generation and creative writing In a recent interview, Sundar Pichai mentioned that Google could have made a better, Ultra-sized Gemini Omni model, but would have had trouble serving it The infrastructure required to serve Ultra-sized models at scale is likely why Google never publicly released models like “Kingfall”

AiBattle

12,010 次观看 • 4 个月前

GPT-5.6 Sol is unbelievably good at creating and editing videos. It can do motion design, product demos, and animations like this one I made by simply giving it a screen recording. GPT 5.6 has the best design taste and significantly outperforms Fable, which relies heavily on repetitive design patterns. To help you experiment with video editing on it, we just launched a collection of 100 ready-to-use skills that show what’s possible and help you get started with video editing using GPT-5.6. These skills can create anything from motion graphics launch videos for your product to a 3B1B-style science explainer video. You can also use them to edit existing videos: add captions, generate motion graphics, create voiceovers, redesign visual styles, translate into new languages, and much more. If you want access to the full library, comment “VIDEO SKILLS” and I’ll share it with you. (You'll have to follow me so I can DM you.)

GPT-5.6 Sol is unbelievably good at creating and editing videos. It can do motion design, product demos, and animations like this one I made by simply giving it a screen recording. GPT 5.6 has the best design taste and significantly outperforms Fable, which relies heavily on repetitive design patterns. To help you experiment with video editing on it, we just launched a collection of 100 ready-to-use skills that show what’s possible and help you get started with video editing using GPT-5.6. These skills can create anything from motion graphics launch videos for your product to a 3B1B-style science explainer video. You can also use them to edit existing videos: add captions, generate motion graphics, create voiceovers, redesign visual styles, translate into new languages, and much more. If you want access to the full library, comment “VIDEO SKILLS” and I’ll share it with you. (You'll have to follow me so I can DM you.)

Akash Anand

518,334 次观看 • 2 个月前

Here’s a real life example of what we have talked about over the last couple of days. Spending all of your time studying limited assets so that you become really good at trading a select 1-2 assets. Live in those charts you’ll get to know them intimately well. My main two are SOL and ETH. But I haven’t traded ETH in a few weeks. It shows on my entry. Lesson in that.

Here’s a real life example of what we have talked about over the last couple of days. Spending all of your time studying limited assets so that you become really good at trading a select 1-2 assets. Live in those charts you’ll get to know them intimately well. My main two are SOL and ETH. But I haven’t traded ETH in a few weeks. It shows on my entry. Lesson in that.

The White Whale

74,221 次观看 • 11 个月前

ANNOUNCEMENT TIME. . TOM A SMITH X SWIM SCHOOL I DONT WANT YOU TO HAVE TO REMEMBER ME FOR LONGER THAN YOU EVER KNEW ME Can I start by saying that this is the best song I’ve ever written. When we play it live it’s had the best reaction I’ve ever had and it’s been requested almost daily since we first played it. It’s also the longest title ever! . I wrote it with a female co-harmony in mind and after a chance meeting at Isle of Wight Festival the amazing Swim School agreed to collaborate. Can I just say that Alice is one of the most creative, innovative performers I have ever met. We ended up with layer upon layer of her incredible vocals that completely dominate the song and take it to another level. What a band man. . . I can’t wait for you to hear it but you’ll need to wait until 30/8.

ANNOUNCEMENT TIME. . TOM A SMITH X SWIM SCHOOL I DONT WANT YOU TO HAVE TO REMEMBER ME FOR LONGER THAN YOU EVER KNEW ME Can I start by saying that this is the best song I’ve ever written. When we play it live it’s had the best reaction I’ve ever had and it’s been requested almost daily since we first played it. It’s also the longest title ever! . I wrote it with a female co-harmony in mind and after a chance meeting at Isle of Wight Festival the amazing Swim School agreed to collaborate. Can I just say that Alice is one of the most creative, innovative performers I have ever met. We ended up with layer upon layer of her incredible vocals that completely dominate the song and take it to another level. What a band man. . . I can’t wait for you to hear it but you’ll need to wait until 30/8.

TOM A SMITH

18,992 次观看 • 2 年前

We are releasing AutoResearchExam, a benchmark on open-ended machine learning and engineering tasks. Our benchmark covers seven research areas including model training, data curation, AI safety and interpretability. In each task, we give agents 24 hours with a CPU or GPU machine to develop and improve their solutions through experiments and feedback. We measure both speed and quality with a combined score. Our benchmark has a unique feature: testing if agents create improvements that hold up on data they never see. We find that AI research agents often overfit as they try to improve. We see an interesting head-to-head comparison at the frontier: Astra starts the strongest and holds the lead for up to 19 hours but Fable 5.1 catches up and gets the top performing spot in the final hours. Qwen3.8 Max, Gemini 3.8 Flash and Grok 4.6 all sit on the cost-performance Pareto frontier, giving strong options at lower API budgets. Anthropic's Opus and Fable retain nearly all their validation performance on hidden tests, with gaps of 1.1% and 2.9%. Astra's improvement over Sol extends to generalization too, with that gap falling from 6.9% to 1.7%. (1/n)

We are releasing AutoResearchExam, a benchmark on open-ended machine learning and engineering tasks. Our benchmark covers seven research areas including model training, data curation, AI safety and interpretability. In each task, we give agents 24 hours with a CPU or GPU machine to develop and improve their solutions through experiments and feedback. We measure both speed and quality with a combined score. Our benchmark has a unique feature: testing if agents create improvements that hold up on data they never see. We find that AI research agents often overfit as they try to improve. We see an interesting head-to-head comparison at the frontier: Astra starts the strongest and holds the lead for up to 19 hours but Fable 5.1 catches up and gets the top performing spot in the final hours. Qwen3.8 Max, Gemini 3.8 Flash and Grok 4.6 all sit on the cost-performance Pareto frontier, giving strong options at lower API budgets. Anthropic's Opus and Fable retain nearly all their validation performance on hidden tests, with gaps of 1.1% and 2.9%. Astra's improvement over Sol extends to generalization too, with that gap falling from 6.9% to 1.7%. (1/n)

Alex Dimakis

2,032,272 次观看 • 24 天前

What’s better than two Agony Aunts? Three! 🤭💃🏽 The amazing and ever-so-funny Hauwa 🌟🌬 is joining us to tackle your dilemmas this week! In this episode, we talk about the realities of running a business in Nigeria, what Hauwa has been up to, how she got into content creation and of course, your dilemmas from catching feelings for your best friend to being hidden by your boyfriend. Tomorrow, be there! 🫵🏾❤️

What’s better than two Agony Aunts? Three! 🤭💃🏽 The amazing and ever-so-funny Hauwa 🌟🌬 is joining us to tackle your dilemmas this week! In this episode, we talk about the realities of running a business in Nigeria, what Hauwa has been up to, how she got into content creation and of course, your dilemmas from catching feelings for your best friend to being hidden by your boyfriend. Tomorrow, be there! 🫵🏾❤️

ISaidWhatISaidPod

119,588 次观看 • 11 天前

Right now, you may not have access to models like GPT‑5.6 Sol, GPT‑4.6 Terra, GPT‑5.6 Luna, Claude Mythos 5, or Claude Fable 5. But you can run something surprisingly powerful today, locally, and completely free. in the next 10 mins on your 8 GB VRAM gaming laptop. Gemma 4 26B A4B QAT (MoE) delivers strong performance on a standard 8 GB VRAM GPU using Ollama, with no API, no usage limits, and no external dependencies. Out of the box, it reaches around 20 tokens per second without any optimizations. Only one command in your terminal: Ollama run gemma4:26b This means: Full offline capability (privacy by default) Zero recurring cost Competitive performance for many real world tasks Fast enough for interactive use on cheap consumer hardware If you're waiting for cutting edge cloud models, you're missing what is already practical today: a capable, local LLM that runs entirely on your own machine.

Right now, you may not have access to models like GPT‑5.6 Sol, GPT‑4.6 Terra, GPT‑5.6 Luna, Claude Mythos 5, or Claude Fable 5. But you can run something surprisingly powerful today, locally, and completely free. in the next 10 mins on your 8 GB VRAM gaming laptop. Gemma 4 26B A4B QAT (MoE) delivers strong performance on a standard 8 GB VRAM GPU using Ollama, with no API, no usage limits, and no external dependencies. Out of the box, it reaches around 20 tokens per second without any optimizations. Only one command in your terminal: Ollama run gemma4:26b This means: Full offline capability (privacy by default) Zero recurring cost Competitive performance for many real world tasks Fast enough for interactive use on cheap consumer hardware If you're waiting for cutting edge cloud models, you're missing what is already practical today: a capable, local LLM that runs entirely on your own machine.

Alok

65,387 次观看 • 3 个月前

I tried explaining to some of our American friends and family members about the sheer magnitude of fireworks being launched throughout the city of Berlin on New Year’s Eve. It is incomprehensible. When we moved here, a fellow American warned me, “Think of the biggest Fourth of July celebration you’ve ever been to and then multiply that by 10 and imagine it lasting 5 hours.” And even that doesn’t begin to describe what New Year’s Eve is like here. 90% of Germany’s fireworks are sold in 3 days, every street becomes a launchpad, and anyone can be a pyrotechnician. 🎆 Here’s our view from last year. And it went on and on for hours.

I tried explaining to some of our American friends and family members about the sheer magnitude of fireworks being launched throughout the city of Berlin on New Year’s Eve. It is incomprehensible. When we moved here, a fellow American warned me, “Think of the biggest Fourth of July celebration you’ve ever been to and then multiply that by 10 and imagine it lasting 5 hours.” And even that doesn’t begin to describe what New Year’s Eve is like here. 90% of Germany’s fireworks are sold in 3 days, every street becomes a launchpad, and anyone can be a pyrotechnician. 🎆 Here’s our view from last year. And it went on and on for hours.

Celia

2,394,324 次观看 • 9 个月前

🗓️ The Portals State of the Union is happening on Sunday, January 28, where we begin to take the wraps off what we’ve been cooking over the last half of 2023. We have been strategically quieter on the timeline, while relentlessly focused on building a self-sustaining, successful platform that thrives with both web2 and web3 audiences. We have been validating secret projects, while competitors seem to be continuing in the wrong direction. The team has been working on the most valuable experiences and products possible now that we have the fastest/best-in-class 3D experience creator out there. Portals has always been a labor of love. It’s our life’s work, and we are not going anywhere. Solana is also here to stay, further establishing itself as a force to be reckoned with and gaining more of the recognition it deserves. It's time for us to put forward more of the on-chain ecosystem projects we have always wanted to do. Rolling out new systems, incentives, and experiences to fortify Portals as one web3's killer apps and top destinations. We have a lot to share, and we can't wait to show you. This year's State of the Union will be delivered in video format, across X, Discord, and in Portals. We'll be dropping more teasers while looking forward to seeing you on January 28!

🗓️ The Portals State of the Union is happening on Sunday, January 28, where we begin to take the wraps off what we’ve been cooking over the last half of 2023. We have been strategically quieter on the timeline, while relentlessly focused on building a self-sustaining, successful platform that thrives with both web2 and web3 audiences. We have been validating secret projects, while competitors seem to be continuing in the wrong direction. The team has been working on the most valuable experiences and products possible now that we have the fastest/best-in-class 3D experience creator out there. Portals has always been a labor of love. It’s our life’s work, and we are not going anywhere. Solana is also here to stay, further establishing itself as a force to be reckoned with and gaining more of the recognition it deserves. It's time for us to put forward more of the on-chain ecosystem projects we have always wanted to do. Rolling out new systems, incentives, and experiences to fortify Portals as one web3's killer apps and top destinations. We have a lot to share, and we can't wait to show you. This year's State of the Union will be delivered in video format, across X, Discord, and in Portals. We'll be dropping more teasers while looking forward to seeing you on January 28!

Portals

37,529 次观看 • 2 年前

Claude "Puzzling" while GPT 5.6 on GOD-MODE just bade a banger that's mindblowing. Here's the exact way to get a site like this, step by step: > open the desktop app, pick Sol, reasoning on High. taste work never goes to small models > drop it 3 sites with motion you love and one line: "reverse-engineer the art direction: mood, typography, pacing, and WHY each animation exists. save it as a style bible" > brief in one paragraph, goal not steps: "[your niche] site, cinematic scroll, every animation has a job. follow the bible" > house rules on top: no template hero, no stock gradients, nothing on the page moves without a reason > now the bar: "a motion designer can't tell this from an agency build." spin up a SECOND 5.6 with fresh context whose only job is to FAIL the build against that bar > /loop overnight: build, grade, close the biggest gap, again. you're asleep for all of it > when the verifier runs out of complaints: tag Sites. live URL, one click, zero hosting The deeper version of every step (the full contract, the house rules, the verifier trick, when Ultra is worth the bill) is in the article below. P.S. send the article to your GPT and tell it "we're doing this tonight".

Claude "Puzzling" while GPT 5.6 on GOD-MODE just bade a banger that's mindblowing. Here's the exact way to get a site like this, step by step: > open the desktop app, pick Sol, reasoning on High. taste work never goes to small models > drop it 3 sites with motion you love and one line: "reverse-engineer the art direction: mood, typography, pacing, and WHY each animation exists. save it as a style bible" > brief in one paragraph, goal not steps: "[your niche] site, cinematic scroll, every animation has a job. follow the bible" > house rules on top: no template hero, no stock gradients, nothing on the page moves without a reason > now the bar: "a motion designer can't tell this from an agency build." spin up a SECOND 5.6 with fresh context whose only job is to FAIL the build against that bar > /loop overnight: build, grade, close the biggest gap, again. you're asleep for all of it > when the verifier runs out of complaints: tag Sites. live URL, one click, zero hosting The deeper version of every step (the full contract, the house rules, the verifier trick, when Ultra is worth the bill) is in the article below. P.S. send the article to your GPT and tell it "we're doing this tonight".

Miraqle

206,624 次观看 • 2 个月前

🚨GAUTAM GAMBHIR MASSIVE STATEMENT ON HIS FUTURE AS INDIA's HEAD COACH🤯 Gautam Gambhir said :🗣️ "I am here to display the results & frankly speaking it has been a roller coaster ride so far. The main goal is always to win major tournaments & team has won Champions trophy & T20 World cup. There will be bad days offcourse, but this is how it goes sonedays. I don't listen to the outside noise & we have set our plans for the 2027 ODI WC. This is our first preference & then you can decide my future. I have been appointed as the coach of Indian cricket team to make it the best team in the world & we have been one of the most consistent teams in white ball cricket for some time now. We are in a transition phase in red ball cricket, and the team will get better with time".

🚨GAUTAM GAMBHIR MASSIVE STATEMENT ON HIS FUTURE AS INDIA's HEAD COACH🤯 Gautam Gambhir said :🗣️ "I am here to display the results & frankly speaking it has been a roller coaster ride so far. The main goal is always to win major tournaments & team has won Champions trophy & T20 World cup. There will be bad days offcourse, but this is how it goes sonedays. I don't listen to the outside noise & we have set our plans for the 2027 ODI WC. This is our first preference & then you can decide my future. I have been appointed as the coach of Indian cricket team to make it the best team in the world & we have been one of the most consistent teams in white ball cricket for some time now. We are in a transition phase in red ball cricket, and the team will get better with time".

Akshat

90,831 次观看 • 2 个月前

Astra (GPT-6) is here!!! I've had early access and tested it like crazy with things like games, code, writing, browser control, presentations and general knowledge work. This is the best model I've ever used. Period. (Incredible demos below in this thread ⬇️) Here's my take on Astra: > It's insanely capable. This feels like a massive improvement, not just an incremental change. This is especially true with zero-shot prompts. > It's all about knowledge work. Slide creation, analysis, writing, and browser control. And oh my...it's so good at browser control. GPT-5.6 was already fantastic at doing things in the browser, Astra is another level and significantly faster. > We're closer than ever (arrived?) at prompt-to-playable game. And I don't just mean only playable, these are actually fun games. I bet if someone with a great eye for games used Astra, they could create a viral game within 1-2 weeks. > Astra is better at writing but not perfect. It removes much of the "AI Smell" we're all familiar with but some stink still survived. > It has a tendency to use the same design colors and look/feel as GPT-5.6 (forrest green anyone?) but it is more steerable in design than previous models. > It's highly steerable in general. A little nudge goes a long way. When I first started using Astra, almost every task I gave it would go for ~30 minutes. I wanted it to keep working. Adding more specifics to a prompt helped greatly with it's ability to work for a long time. > Astra's 3D understanding is unmatched. 3D asset creation was consistent and easy and its spacial awareness while building complex 3D worlds blew me away. I'm still getting familiar with Astra but this will now be my go-to model for any difficult work I have. Check out the demos below: 👇

Astra (GPT-6) is here!!! I've had early access and tested it like crazy with things like games, code, writing, browser control, presentations and general knowledge work. This is the best model I've ever used. Period. (Incredible demos below in this thread ⬇️) Here's my take on Astra: > It's insanely capable. This feels like a massive improvement, not just an incremental change. This is especially true with zero-shot prompts. > It's all about knowledge work. Slide creation, analysis, writing, and browser control. And oh my...it's so good at browser control. GPT-5.6 was already fantastic at doing things in the browser, Astra is another level and significantly faster. > We're closer than ever (arrived?) at prompt-to-playable game. And I don't just mean only playable, these are actually fun games. I bet if someone with a great eye for games used Astra, they could create a viral game within 1-2 weeks. > Astra is better at writing but not perfect. It removes much of the "AI Smell" we're all familiar with but some stink still survived. > It has a tendency to use the same design colors and look/feel as GPT-5.6 (forrest green anyone?) but it is more steerable in design than previous models. > It's highly steerable in general. A little nudge goes a long way. When I first started using Astra, almost every task I gave it would go for ~30 minutes. I wanted it to keep working. Adding more specifics to a prompt helped greatly with it's ability to work for a long time. > Astra's 3D understanding is unmatched. 3D asset creation was consistent and easy and its spacial awareness while building complex 3D worlds blew me away. I'm still getting familiar with Astra but this will now be my go-to model for any difficult work I have. Check out the demos below: 👇

Matthew Berman

1,907,120 次观看 • 1 个月前

i generated this video using two blurry pics of myself from two years ago at griffith observatory with the MiniMax H3 model. it's surprisingly natural and sounds really close to how i actually speak. i haven't followed video generation models enough in the past year, but i'm really curious about the tech behind them. now they can even generate high-fidelity voice, accurate text, and editable video. i still remember the days when we were shocked by gpt-4o and the sora sample video. it's been a long way since then. #MiniMaxH3 MiniMax Design (H3)

i generated this video using two blurry pics of myself from two years ago at griffith observatory with the MiniMax H3 model. it's surprisingly natural and sounds really close to how i actually speak. i haven't followed video generation models enough in the past year, but i'm really curious about the tech behind them. now they can even generate high-fidelity voice, accurate text, and editable video. i still remember the days when we were shocked by gpt-4o and the sora sample video. it's been a long way since then. #MiniMaxH3 MiniMax Design (H3)

Asuka Zheng🎀

20,908 次观看 • 2 个月前

Today, we're launching the Runway Game Worlds Beta. Over the last few months, we have been working on research and products that are moving us closer toward a future where you will be able to explore any character, story or world in real time. While generating the pixels of these experiences is one aspect of this new frontier, another is the need for novel mechanics and interfaces. From how stories unfold to how your choices affect the worlds you’re simulating. Today’s beta release marks a first step in this direction, learn more below. (1/5)

Today, we're launching the Runway Game Worlds Beta. Over the last few months, we have been working on research and products that are moving us closer toward a future where you will be able to explore any character, story or world in real time. While generating the pixels of these experiences is one aspect of this new frontier, another is the need for novel mechanics and interfaces. From how stories unfold to how your choices affect the worlds you’re simulating. Today’s beta release marks a first step in this direction, learn more below. (1/5)

Runway

118,827 次观看 • 1 年前

This is my "feel the AGI" moment: I used GPT-5.6 Sol to train my own autocorrect model that outperforms GPT-5.6 Sol (wtf??) I have no ML background. I have no idea what I'm doing. I just kept pushing Sol until it spat out a SOTA model. And I spent $0. The motivation: Years of talking to AI have made me terrible at typing. Rather than fix my skill issue, I decided to throw more AI at it. My idea was: instead of autocorrect that interrupts my flow, I want to type fast with mistakes and have AI clean it up after. I wanted the smallest local model possible, for speed, for battery life, for science! So I decided to train my own. Inspired by Andrej Karpathy’s autoresearch, I ran Codex /goal with this setup: pick an experiment, try it, record the results to a doc, throw it out if it fails, and plan the next experiment without repeating failures. I gave a few examples that had to pass, tight latency targets, and let it run. Sol did some amazing things. First, it scanned benchmarks and shortlisted base models: Qwen 3.5, Gemma 4, Liquid LFM 2.5. It found a dataset on HuggingFace for typed text. Then it built a simulator for fingers striking a Mac keyboard, modeling the physical layout with a Gaussian distribution around each key. It simulated striking the wrong key, wrong order, fat-fingering, etc. With the models + data + simulator, it fine-tuned using MLX right on my MacBook. It had a working prototype within an hour! But accuracy was pretty poor. — Problem 1: Tokenization Sol read papers, ran tests, and identified that the tokenizer was the bottleneck. Tokenization makes typos hard for the model to see, so it memorizes mappings instead of using its language priors. Sol tried ByT5, Google’s tokenizer-free byte-level LLM. This made a big improvement, but the model is old and lacked the knowledge needed to reach Sol performance. Sol dug deeper and realized a tokenizer-free model isn’t needed; instead, it used T5Gemma, an encoder-decoder model. This can understand the input deeply before producing output, and furthermore, Sol could post-train the encoder to improve performance. This gave a much higher ceiling. — Problem 2: Loss function Now the model was correcting some typos perfectly, but ignoring most. Sol realized that standard cross-entropy loss was teaching the model to avoid edits, because the vast majority of characters in the training data were left unmodified. The fix was wild: Sol wrote a custom loss function that byte-aligns the source and target strings, uses a dynamic programming algorithm to compute the minimum edits between the two, then weights correct edits much higher than copies. After a lot of tuning, this dramatically improved accuracy. — Problem 3: Autoregression One failure mode remained: if the model made a mistake, it couldn’t backtrack. It could only predict the next token. Teaching it to “think” like a reasoning model would solve this, but would be far too slow. Sol found a beautiful solution: instead of greedily predicting the next token, beam search over all possibilities. This parallelizes the exploration instead of one linear chain-of-thought. At the end, choose the path with highest cumulative log probability. This worked great, but made the experience worse, since the user wouldn’t see progress until the whole search was done. To fix this, Sol made a clever observation: after each search step, the longest common prefix among surviving branches is guaranteed to appear in the final result, so it can be displayed immediately. As the search progresses, weaker paths are dropped and the prefix grows, so the user sees continuous progress. Sol built all this as a custom MLX pipeline that does the parallel decoding on the MacBook GPU, with just ~40ms TTFT. It’s crazy fast and entirely local. — Final eval (error reduction rate, higher is better): - Apple autocorrect: 49.66% - GPT-5.6 Luna: 82.47% - GPT-5.6 Terra: 87.64% - GPT-5.6 Sol: 90.56% - Our model (1.7B): 91.02% Final cost: - 1 quota reset (thanks Tibo) - $0 (And yes, I verified there's no cheating. In fact, we test words scrubbed from the training data to prove the model isn’t memorizing) There were a ton more details and tangents I could write about: contrastive learning, GRPO, DPO, dynamic masking, and more. Sol is a fascinating and creative model. It blew my mind so many times. Don’t let a lack of experience stop you: Sol makes AI experiments accessible to anyone!

This is my "feel the AGI" moment: I used GPT-5.6 Sol to train my own autocorrect model that outperforms GPT-5.6 Sol (wtf??) I have no ML background. I have no idea what I'm doing. I just kept pushing Sol until it spat out a SOTA model. And I spent $0. The motivation: Years of talking to AI have made me terrible at typing. Rather than fix my skill issue, I decided to throw more AI at it. My idea was: instead of autocorrect that interrupts my flow, I want to type fast with mistakes and have AI clean it up after. I wanted the smallest local model possible, for speed, for battery life, for science! So I decided to train my own. Inspired by Andrej Karpathy’s autoresearch, I ran Codex /goal with this setup: pick an experiment, try it, record the results to a doc, throw it out if it fails, and plan the next experiment without repeating failures. I gave a few examples that had to pass, tight latency targets, and let it run. Sol did some amazing things. First, it scanned benchmarks and shortlisted base models: Qwen 3.5, Gemma 4, Liquid LFM 2.5. It found a dataset on HuggingFace for typed text. Then it built a simulator for fingers striking a Mac keyboard, modeling the physical layout with a Gaussian distribution around each key. It simulated striking the wrong key, wrong order, fat-fingering, etc. With the models + data + simulator, it fine-tuned using MLX right on my MacBook. It had a working prototype within an hour! But accuracy was pretty poor. — Problem 1: Tokenization Sol read papers, ran tests, and identified that the tokenizer was the bottleneck. Tokenization makes typos hard for the model to see, so it memorizes mappings instead of using its language priors. Sol tried ByT5, Google’s tokenizer-free byte-level LLM. This made a big improvement, but the model is old and lacked the knowledge needed to reach Sol performance. Sol dug deeper and realized a tokenizer-free model isn’t needed; instead, it used T5Gemma, an encoder-decoder model. This can understand the input deeply before producing output, and furthermore, Sol could post-train the encoder to improve performance. This gave a much higher ceiling. — Problem 2: Loss function Now the model was correcting some typos perfectly, but ignoring most. Sol realized that standard cross-entropy loss was teaching the model to avoid edits, because the vast majority of characters in the training data were left unmodified. The fix was wild: Sol wrote a custom loss function that byte-aligns the source and target strings, uses a dynamic programming algorithm to compute the minimum edits between the two, then weights correct edits much higher than copies. After a lot of tuning, this dramatically improved accuracy. — Problem 3: Autoregression One failure mode remained: if the model made a mistake, it couldn’t backtrack. It could only predict the next token. Teaching it to “think” like a reasoning model would solve this, but would be far too slow. Sol found a beautiful solution: instead of greedily predicting the next token, beam search over all possibilities. This parallelizes the exploration instead of one linear chain-of-thought. At the end, choose the path with highest cumulative log probability. This worked great, but made the experience worse, since the user wouldn’t see progress until the whole search was done. To fix this, Sol made a clever observation: after each search step, the longest common prefix among surviving branches is guaranteed to appear in the final result, so it can be displayed immediately. As the search progresses, weaker paths are dropped and the prefix grows, so the user sees continuous progress. Sol built all this as a custom MLX pipeline that does the parallel decoding on the MacBook GPU, with just ~40ms TTFT. It’s crazy fast and entirely local. — Final eval (error reduction rate, higher is better): - Apple autocorrect: 49.66% - GPT-5.6 Luna: 82.47% - GPT-5.6 Terra: 87.64% - GPT-5.6 Sol: 90.56% - Our model (1.7B): 91.02% Final cost: - 1 quota reset (thanks Tibo) - $0 (And yes, I verified there's no cheating. In fact, we test words scrubbed from the training data to prove the model isn’t memorizing) There were a ton more details and tangents I could write about: contrastive learning, GRPO, DPO, dynamic masking, and more. Sol is a fascinating and creative model. It blew my mind so many times. Don’t let a lack of experience stop you: Sol makes AI experiments accessible to anyone!

Anshu

179,451 次观看 • 2 个月前

“The Last Meatball” Animation: jboogx.creative & enigmatic_e Movement: jboogx.creative & enigmatic_e Reference Imagery: FLUX on Civitai Jboogx and I have known each other for the better part of the last 12 months. I think we've both been inspired by each other's work and dedication to exploring all the ways we can push these tools to their limits for VFX and animation. We didn't want to force a collaboration, but recently, Jboogx was inspired by the work I've been doing and how I incorporate myself into my animations. This idea is a playful take on 'Atlas carrying the boulder,' but with a twist—let's make Atlas out of spaghetti and the boulder a meatball, to fit more with what Jboogx has been doing on the food side of things lately (shoutout to our good friend James Gerde -@gerdegotit @gerdedoesit @gerdemadeit, who did the first spaghetti animation). Since I was going to record myself, it wouldn't have been right if Jboogx didn't as well! This is just a taste of what’s possible. I'm located in Germany, and Jboogx is in Hawaii. A world apart, and we pulled this together in 48 hours. BTS coming soon 😉

“The Last Meatball” Animation: jboogx.creative & enigmatic_e Movement: jboogx.creative & enigmatic_e Reference Imagery: FLUX on Civitai Jboogx and I have known each other for the better part of the last 12 months. I think we've both been inspired by each other's work and dedication to exploring all the ways we can push these tools to their limits for VFX and animation. We didn't want to force a collaboration, but recently, Jboogx was inspired by the work I've been doing and how I incorporate myself into my animations. This idea is a playful take on 'Atlas carrying the boulder,' but with a twist—let's make Atlas out of spaghetti and the boulder a meatball, to fit more with what Jboogx has been doing on the food side of things lately (shoutout to our good friend James Gerde -@gerdegotit @gerdedoesit @gerdemadeit, who did the first spaghetti animation). Since I was going to record myself, it wouldn't have been right if Jboogx didn't as well! This is just a taste of what’s possible. I'm located in Germany, and Jboogx is in Hawaii. A world apart, and we pulled this together in 48 hours. BTS coming soon 😉

enigmatic_e

11,422 次观看 • 2 年前